AMD stock jumped nearly 4% on Tuesday after Raymond James upgraded Advanced Micro Devices to Strong Buy from Outperform and lifted its price target to $641 from $565, arguing that agentic AI could create a much larger server CPU opportunity than Wall Street currently appreciates. The firm believes CPUs will become increasingly critical as AI agents coordinate accelerators, retrieve data and manage applications, giving AMD another growth engine beyond its Instinct GPUs. The new target implies roughly 40% upside from AMD’s Monday close near $456, putting the spotlight back on whether AMD can keep taking data-center share from Intel while narrowing Nvidia’s lead in AI infrastructure.
The upgrade lands at an unusually important moment for AMD. The company just reported record quarterly revenue of $11.54 billion, with Data Center sales more than doubling to $6.7 billion, and management expects that business to accelerate again in the second half of 2026. At the same time, AMD is rolling out its new Helios rack-scale AI architecture, 6th Gen EPYC CPUs and MI400-series accelerators while signing major customers including Microsoft, Meta, OpenAI and Anthropic. The bullish thesis is therefore no longer simply that AMD might participate in AI. It is that the company could become one of the few vendors able to sell a complete AI infrastructure stack spanning CPUs, GPUs, networking and software.
Raymond James Says Agentic AI Could Transform the Server CPU Market
Raymond James analyst Simon Leopold raised AMD to Strong Buy and increased the price target to $641 from $565, saying the firm is extending its “AI factory” framework beyond accelerators and into server CPUs. Leopold forecasts the broader server CPU market could grow at a roughly 44% compound annual rate to about $201 billion by 2030, driven by traditional data-center demand, CPUs required to host and coordinate accelerators and a new wave of agentic AI workloads.
The agentic AI angle is important because much of the market’s AI attention has centered on GPUs. Large language models require accelerators for training and inference, but autonomous AI agents do considerably more than generate text. They may continuously query databases, call software tools, coordinate multiple models, manage memory, execute workflows and interact with enterprise applications. Those tasks create significant demand for general-purpose processing alongside GPUs. Raymond James argues that AMD is particularly well positioned because EPYC server processors are already taking share in conventional data centers just as AI creates an entirely new layer of CPU demand. In Leopold’s view, AMD offers one of the strongest combinations of direct earnings leverage, favorable data-center exposure and continuing market-share gains.
Latest Earnings Give the Upgrade Real Fundamental Support
The upgrade is arriving on top of unusually strong operating results. AMD reported second-quarter revenue of $11.54 billion, up 50% year over year, while non-GAAP operating income jumped 245% to $3.09 billion and adjusted EPS rose to $1.66 from $0.48 a year earlier. Most importantly, Data Center revenue reached $6.7 billion, up 107%, accounting for 58% of total company sales. AMD said the increase reflected strong demand for both EPYC processors and Instinct accelerators.
Related articles
Management is also telling investors that growth is not peaking yet. AMD expects third-quarter revenue of approximately $13 billion, plus or minus $300 million, implying about 41% year-over-year growth and another 13% sequential increase. Non-GAAP gross margin is expected at roughly 56%. CFO Jean Hu said Data Center sales should accelerate during the second half of 2026, while CEO Lisa Su pointed to rising EPYC demand, scaling Instinct deployments and the beginning of the Helios ramp. That guidance makes the Raymond James call more than a thematic AI argument: AMD is already showing the revenue and margin expansion investors would expect if it is successfully gaining share.
EPYC Could Be the Underappreciated Part
Wall Street often treats AMD primarily as Nvidia’s challenger in AI accelerators, but Raymond James is emphasizing a business where AMD arguably has a more established competitive position: server CPUs. AMD’s EPYC franchise has spent years taking share from Intel, and recent market research indicates those gains continued during the second quarter. Mercury Research data cited by Tom’s Hardware showed AMD reaching record x86 share across multiple segments, with server strength particularly notable even as Intel remained the overall market leader.
That becomes more financially important if agentic AI increases the amount of CPU compute required per GPU cluster. Every accelerator rack still needs CPUs to orchestrate workloads, feed data to accelerators and handle applications that do not belong on GPUs. AMD’s newly introduced 6th Gen EPYC processors are explicitly designed for agentic AI, general-purpose and enterprise workloads, and the company says they provide the performance and memory bandwidth required to keep accelerator systems fed efficiently. If Raymond James is right that AI expands the CPU market rather than simply shifting spending toward GPUs, AMD could benefit twice: once through EPYC share gains and again through growing Instinct deployments.
Helios Is AMD’s Biggest Attempt Yet to Challenge Nvidia
The second major part of the bull case is Helios. AMD unveiled the rack-scale architecture in July alongside the MI400 family and 6th Gen EPYC CPUs, moving beyond individual chip sales toward complete AI systems. AMD says Helios can deliver up to 30% more inference tokens per dollar than competing solutions under its internal testing assumptions, although those performance claims will ultimately need to be validated across real customer deployments.
The strategic importance is clear. Nvidia’s advantage is not just the performance of individual GPUs; it comes from selling integrated racks, networking, CPUs, accelerators and a mature software ecosystem. Helios is AMD’s most serious attempt to compete at that full-system level. OpenAI expects to bring Helios online beginning in the fourth quarter of 2026, Microsoft plans to deploy it at scale on Azure, and Meta is testing and validating the architecture for major deployments. If these customers move beyond pilots and into sustained multi-gigawatt purchases, AMD’s addressable market could expand dramatically beyond the role of secondary GPU supplier.
Meta and Anthropic Give Something Wall Street Has Wanted: Scale
Large customer commitments are beginning to make the story more tangible. In February, AMD and Meta announced a multi-year agreement covering up to 6 gigawatts of AMD Instinct GPUs, with shipments supporting the first gigawatt scheduled to begin in the second half of 2026. The deal aligns GPU, CPU, system and software roadmaps and uses a custom MI450-based accelerator alongside EPYC processors and ROCm software.
Anthropic followed in July with an agreement to deploy up to 2 gigawatts of MI450-series GPUs in Helios systems, with the first gigawatt expected to begin deployment in the first half of 2027. AMD also committed to invest up to $5 billion in Anthropic as part of the strategic relationship. These partnerships are important because one of the biggest historical concerns around AMD’s AI business has been whether large customers would commit at Nvidia-like scale. Multi-gigawatt agreements do not guarantee flawless execution or margins, but they provide much stronger revenue visibility than isolated benchmark wins.
Nvidia Is Still the Benchmark — and the Biggest Risk
None of this means AMD has displaced Nvidia. Nvidia remains the dominant supplier of AI accelerators and continues to benefit from a mature CUDA software ecosystem, enormous scale and rapid product cycles. Nvidia’s upcoming Vera Rubin platform is expected to ship this autumn, meaning AMD will be competing against a moving target rather than a static incumbent. Reuters reported Tuesday that investors are watching Rubin closely as Nvidia faces increasing competition from AMD, Intel and custom silicon while the AI spending boom itself comes under greater financing scrutiny.
For AMD stock, this is the biggest strategic risk behind Raymond James’s bullish target. AMD does not merely need Helios and MI400 to perform well; it needs enough customers to deploy those systems at scale while maintaining competitive economics and improving ROCm software support. Nvidia can also respond with pricing, accelerated product launches or tighter integration across its platform. The bullish case therefore assumes a rapidly expanding AI market in which AMD can win meaningful share without needing to defeat Nvidia outright. If the total market grows quickly enough, second place can still be extraordinarily profitable.
Valuation Makes Execution More Important Than Ever
The upgrade also arrives after an enormous run in AMD shares. The stock has more than doubled in 2026 and has climbed more than 170% over the past 12 months, according to market data cited by Stocktwits. Raymond James’s $641 target still implies substantial upside, but the market is already assigning AMD a much richer valuation than it carried before its AI revenue began accelerating.
That creates a higher bar for every quarterly report. Investors will increasingly expect Data Center revenue to keep growing rapidly, Helios deployments to arrive on schedule and gross margins to improve as the product mix shifts toward higher-value AI systems. Any delay in MI450 availability, packaging constraints, weaker customer deployments or unexpectedly aggressive pricing could pressure the stock even if AMD continues gaining share. The company has announced more than $10 billion of investments across Taiwan’s technology ecosystem to expand supply and advanced packaging capability, underlining how much capital and execution are required to support the next phase of growth.
Outlook: What Stock Investors Should Watch Next
Raymond James’s upgrade strengthens the argument that investors should think about AMD as more than a GPU challenger. If agentic AI substantially increases server CPU demand, AMD could monetize the AI boom through EPYC even before accounting for its accelerating Instinct and Helios businesses. The latest numbers support that case: Data Center revenue more than doubled last quarter, total revenue reached a record $11.5 billion and management expects another step higher to roughly $13 billion in Q3.
The next milestones are straightforward. Investors should watch whether Meta, Microsoft, OpenAI and Anthropic move Helios deployments forward on schedule, whether EPYC continues taking server share, whether AMD sustains its 56% area gross margins and whether 2027 AI revenue visibility improves as MI450 ramps. Nvidia’s results and Rubin launch will also provide an important benchmark for how fast the broader market is expanding.
Raymond James now sees $641 as a reasonable destination for AMD stock because agentic AI could make CPUs just as strategically important as the accelerators dominating today’s headlines. The market has already rewarded AMD for becoming a credible AI competitor. The next leg higher will depend on proving it can turn that credibility into something much bigger: a full-scale AI infrastructure franchise.
